Multivariate factor analysis of heavy minerals concentrate from Athabasca oil sands tailings by X‐ray photoelectron spectroscopy

Abstract

Multivariate factor analysis of X‐ray photoelectron spectroscopy data acquired from Athabasca oil sands heavy minerals concentrate was used to identify the primary mineral components and their physical associations. Using large‐area spectroscopy, a principal components analysis (PCA) and Varimax rotation of the PCA spectral loadings matrix demonstrated that the most significant factors index the mineral chemistry by virtue of the within‐factor spectral correlations. Analysis of the Varimax rotated factor scores indicated the physical character of several mineral associations. Emphasis is placed on the high‐value materials, namely, zircon and the titanium‐bearing minerals. In spectral imaging mode, Varimax rotation in the spatial domain applied to a PCA noise‐reduced data reconstruction was used to render component images illustrating the spatial distribution of selected mineral chemistries. The component images also revealed evidence of surface species consistent with pyrite weathering. Data are supported with optical microscopy and energy‐dispersive X‐ray spectroscopy. Our work demonstrates the utility of multivariate spectroscopic techniques in the analysis of complex mineral chemistry. © 2013 National Research Council Canada and John Wiley & Sons Ltd.